The inclusion criteria were:

Primary newly diagnosed WHO grade 4 astrocytoma adult patients (age over 18 years) who underwent surgery. Gross total resections (GTR) and Near Total Resection (NTR) were defined as no residual tumor enhancement and an extent of resection of more than 95% of the initial enhancing volume, respectively. Patients were treated with systemic temozolomide according to the Stupp protocol. Tumor progression was defined according to the modified Response assessment in neuro-oncology criteria (RANO). All the patients in our collection had primary glioblastomas. They were all newly diagnosed, with the exception of two patients who had undergone previous surgery and chemo/radiotherapy.

The exclusion criteria were:

Other histopathological diagnoses, patients in which it was impossible to establish the diagnosis of progression vs. pseudo-progression, missing MRI sequences, and poor-quality MRI scans due to the presence of artifacts.

The dataset includes clinical and pathological information:

Age, Sex, preoperative and postoperative Karnofsky performance score, Overall survival, Progression-free survival, percentage of the extent of resection of enhancing tumor, systemic therapy received, details of RT received (dose, technique, number of fractions, isodose), IDH status, ATRX mutation, and Ki-67 index, size of enhancing tumor recurrence.

Note:  in the Clinical data file, In this dataset, some patients were still alive at the end of the data collection period, and their survival times are not yet known. These patients are considered right-censored = yes. In survival analysis, ‘right censoring’ occurs when the event of interest has not occurred for some study participants by the end of the study period or by the time the data was collected. This means that the survival time of censored participants is not fully observed or known.

The dataset includes the segmentations of the enhancing tumor, necrosis, and peritumoral region from the pre-postoperative and follow-up studies that experts have manually corrected. The dataset represents a sample of unique characteristics by including patients with an extent of resection of > 95 % of the enhancing tumor.

MRI acquisition protocol:

The dataset includes T1-weighted (T1w), T2-weighted (T2w), FLAIR (Fluid attenuated inversion recovery), T1w contrast-enhanced (T1ce) sequences, and diffusion-weighted imaging-derived apparent diffusion coefficient (ADC) maps from 3 scanners at 2 centers. Please see the manuscript, Supplementary Table 1: MRI acquisition parameters, for detail.

MRI scans underwent preprocessing using a specific pipeline.

• This process involved converting DICOM to NiFTI using the dicom2niix tool (https://github.com/rordenlab/dcm2niix/releases/tag/v1.0.20220720),
• registering T1ce scans for each subject to the SRI24 anatomical atlas space using FLIRT (FMRIB’s Linear Image Registration Tool) (https://fsl.fmrib.ox.ac.uk/fsl/fslwiki/FSL),
• coregistering T1w, T2w, FLAIR scans, and ADC maps to the transformed T1ce scan,
• extracting the brain from all co-registered scans with Synthstrip (https://surfer.nmr.mgh.harvard.edu/docs/synthstrip/), and
• normalizing intensity using the normalization tools included in the Cancer Imaging Phenomics Toolkit (CaPTk) (https://www.nitrc.org/projects/captk/).
• Computer-assisted segmentations were created using Deep-Medic (https://github.com/deepmedic/deepmedic) from the preprocessed images at each time point, resulting in three labels: 1) necrosis, 2) peritumoral signal alteration (including edema and non-enhancing tumor), and 3) enhancing tumor. Two expert neurosurgeons reviewed and manually corrected all segmentations.

Publications by the Dataset Authors

The authors recommended the following as the best source of additional information about this dataset:

No other publications were recommended by dataset authors.

Research Community Publications

TCIA maintains a list of publications that leveraged this dataset. If you have a manuscript you’d like to add please contact TCIA’s Helpdesk.